Accidentally delete project

I just accidentally delete my roboflow project that i’m still working on. Did roboflow have some ways to recover a deleted project?
Thank you

Hi @jessicawong - I just replied to your email.

I’m adding the response here, for visibility for others:

“Unfortunately, projects are permanently deleted after confirming the project name in the deletion window (typing it to confirm, or copy/pasting the name to confirm), and selecting “Delete.”


I’ll also ask my team if there’s any way to locate and restore it. I just don’t know of one at this time”, as the deletion process offers double confirmation. This is also to ensure we don’t keep data that people do not want hosted on the platform (to respect their data privacy rights).

I’m asking my team to learn if there is another way to recover it.

Okay, thank you so much for your help

I’ve received an answer - and unfortunately, we do not have a way to recover it, as we purge the record after the double-confirmation for deletion.

Are there past dataset versions you exported and saved? Or the original image files - are they still available to you?

If there is a generated version with augmentations, or other preprocessing, other than Auto-Orient, that you exported: these are fine to upload and continue working with without compromising your work or your model.

If the generated version you saved and exported is only of augmented or resized versions, then you’d need to consider that factor for the project, as you’d no longer have truly “original” images, but images that are augmented and/or preprocessed (manipulated images, that will not fit with the rest of your “new” images).

Otherwise, the option to upload the original project is the only one left, unfortunately. And this would require re-annotating the images if none of them are labeled.

I will add a few training credits to your project so you can leverage more Label Assist checkpoints after using Roboflow Train to get them labeled faster:

NOTE - I recommend labeling a subset, generating a version for it, and training for your Label Assist checkpoint

Alright, thank you so much for your help mister Mohamed, it means a lot. Have a wonderful day.

1 Like